Error correction coding apparatus

Abstract

An error correction coding apparatus is disposed to generate a low-density parity-check code 16 from an input information sequence 15 by using a low-density parity-check matrix which satisfies a predetermined weight distribution, and includes a low-density parity-check matrix output means 13 for forming the above-mentioned low-density parity-check matrix by continuously arranging a number of rows in each of which the same number of cyclic-permutation matrices as the row weight are arranged, the number of rows satisfying the above-mentioned predetermined weight distribution, and then gradually increasing or decreasing the row weight, and for outputting the above-mentioned low-density parity-check matrix.

The invention claimed is: 
     
       1. An error correction coding apparatus which generates a low-density parity-check code from an input information sequence by using a low-density parity-check matrix which satisfies a predetermined weight distribution, said error correction coding apparatus comprising:
 a low-density parity-check matrix storage unit for storing information used for forming said low-density parity-check matrix; and 
 a coding unit for encoding said input information sequence according to the low-density parity-check matrix, 
 wherein said low-density parity-check matrix comprises at least a first set of adjacent rows corresponding to a first row weight, and a second set of adjacent rows corresponding to a second row weight, said first set of adjacent rows being adjacent to said second set of adjacent rows in said low-density parity-check matrix, 
 wherein each of the rows of said first set of adjacent rows contains a same number of cyclic-permutation matrices corresponding to said first row weight, 
 wherein each of the rows of said second set of adjacent contains a same number of cyclic-permutation matrices corresponding to said second row weight, 
 wherein the number of rows of said first set of adjacent rows and the number of rows of said second set of adjacent rows satisfy said predetermined weight distribution, and 
 wherein said number of cyclic-permutation matrices corresponding to said second row weight is determined by increasing or decreasing said number of cyclic-permutation matrices corresponding to said first row weight by one. 
 
     
     
       2. The error correction coding apparatus according to  claim 1 , said low-density parity-check matrix storage unit stores the respective first rows of said first and second sets of adjacent rows in said low-density parity-check matrix in advance of encoding said input information sequence, and outputs said stored rows to said coding unit, and
 wherein said coding unit reconstructs the rows of said first and second sets of adjacent rows other than said stored rows outputted from said low-density parity-check matrix storage unit by carrying out cyclic shifts of said stored rows outputted from said low-density parity-check matrix storage unit, and for generating the low-density parity-check code from the input information sequence by using the reconstructed rows and the stored rows outputted from said low-density parity-check matrix storage unit. 
 
     
     
       3. The error correction coding apparatus according to  claim 1 , wherein said low-density parity-check matrix storage unit uses, as said predetermined weight distribution, a weight distribution according to a Poisson distribution. 
     
     
       4. The error correction coding apparatus according to  claim 1 ,
 wherein a number of rows of a submatrix which consists of rows of a weight w (w is a natural number equal to or larger 3) is expressed as R(w), an information length is expressed as K (K is a natural number), and a row position in said submatrix is expressed as i (0<=i<=R(w)−1), and 
 wherein the low-density parity-check matrix storage unit sets elements at an ith column and a (K−R(w)+1+i)th column of a row at the row position i of said submatrix to 1, arranges cyclic-permutation matrices in said submatrix so that (w−2) 1s are distributed at almost equal intervals between the ith column and the (K−R(w)+1+i)th column, and outputs said submatrix to said coding unit.